A Novel despeckling framework for Ultra-high Resolution SAR Images Based on Complex Generalized Gaussian distribution

Abstract:
The complex noise behavior and rich details make it very hard to achieve effective noise reduction and detail preservation at the same time for ultra-high resolution (UHR) SAR images. Therefore, an extendable despeckling framework that employs the complex generalized Gaussian distribution (CGGD) to distinguish structural areas and distributed areas and filters them separately to solve the aforementioned problem is proposed. As an example, a new filter derived from this framework is presented and compared with four state-of-the-art filters. Results demonstrate that the new filter outperforms the other ones and can simultaneously realize excellent structural detail preservation and effective noise reduction.